Comments on: Secondary Axes in Charts

By: Secondary axes in Excel 2007 « Newton Excel Bach, not (just) an Excel Blog

Secondary axes in Excel 2007 « Newton Excel Bach, not (just) an Excel Blog — Sat, 20 Feb 2010 03:01:51 +0000

[...] is another matter, but for some ideas on what you can do with secondary axes, have a look at the Jon Peltier Blog (see both the linked post and the “related posts” listed at the bottom for some useful [...]

By: Jon Peltier

Jon Peltier — Wed, 26 Mar 2008 03:14:20 +0000

Hi Lee -

You bring up a very good point. Having worked for many years as an engineer in the US, I’ve had to do my share of work in mixed units. Pounds and kilograms, meters and feet, Fahrenheit and Celsius, you name it. However, this is a different situation than Stephen laid out in his essay. He was critical of multiple series plotted on axes with unrelated setss of units ($ sales vs. units sold, for example). You are normalizing your data to cost per energy content, so you have a single value scale, with multiple ways to represent that scale depending on the type of fuel.

That is like a chart that shows daily temperatures, with Celsius on one axxiss and Fahrenheit on the other. The reader can look at the data points, and look left or right to determine the temperature in whatever units are most suitable. In fact, while waiting for this page to load so I could approve your comment (stupid hosting company), I set up some charts illustrating just this issue, and showing how to set up a “nice” converted axis. Stay tuned, it may be up by tomorrow at this time, if I don’t fall asleep waiting for the server.

Server update: They have finally admitted that this new “platform” they’ve migrated me onto is woefully underequipped to handle lots of MySQL requests. By the weekend they plan to have five additional MySQL servers on line, and things should again be hunky dory. I can’t believe their prediction of server requirements was so far off. I worked on a project a couple years ago where we predicted just this kind of server loading (I had to lern queuing theory and all that), and we got to where our predictions were within a factor of 2. Well, in fairness, I guess my hosting company was within a factor of ZERO.

By: Lee

Lee — Wed, 26 Mar 2008 00:41:49 +0000

I am an analyst in the power business, and I am regularly producing charts comparing prices across multiple fuels – natural gas, fuel oil, coal. Each fuel has its own price units – $/mmbtu, $/gal, $/ton – that my audience expects to see. I use the secondary axis to appropriately scale a second fuel price relative to another on the same chart. Since the fuels are substitute commodities, the relationships and correlations through long periods of time hold true, and can be demonstrated well on a single chart.

Because of the simple conversion factors ($/mmbtu = 7.21*$/gal = 1/25*$/ton), I could build these charts with one axis, covert the data and then label the points with the original prices. But it seems so much easier to have the secondary axis defined, the labels are the data, and my work is straight-forward.

I do agree that other data might be less appropriate for this method, but similar type items with scalar relationships seem to make sense. I have tried to use the Normalized method, but that back fires with more confusion.

What gets me is when the scaling of two axes is not tied together when there is a simple defined relationship between the points.